SheetgoDevelopers
Sheetgo developer docs

Build on Sheetgo with the REST API or the MCP server

Sheetgo gives you two ways in. The REST API manages Sheetgo itself: workflows, automations, triggers, forms, dashboards and runs. The MCP server lets an AI assistant work directly in your Google Sheets, Drive, Docs, PDFs and BigQuery. Both use the same Sheetgo API key. This page explains both, helps you choose, and links to the full reference.

REST API

74 HTTP endpoints in 13 groups, described by OpenAPI. For scripts, back ends, integrations and agents that call HTTP.

Base URL: https://dev-api.sheetgo.com/rest

Open the API docs →

MCP server

35 tools over the Model Context Protocol (Streamable HTTP). For AI assistants such as Claude, Cursor or VS Code agents.

Endpoint: https://dev-api.sheetgo.com/mcp

Open the MCP docs →

For AI agents

Plain, server-rendered HTML pages, one per topic, with an index for LLMs.

Index for LLMs: llms.txt

How to read these docs →
Site map

The pages

MCP serverConnect an AI assistant to Google Sheets, Drive, Docs, PDFs and BigQuery: setup, errors, permissions and all 35 tools. REST APIAuth, conventions, errors, quickstart and the index of all 74 endpoints. API · workflowsA workflow is an automated process that connects your spreadsheets and files and moves data between them — on a schedule or on demand — so a task… API · automationsAn automation (also called a connection) is a single data transfer inside a workflow: it pulls data from one or more sources, optionally… API · triggersA trigger defines what makes a workflow run automatically, so you set it once instead of running the workflow by hand. API · dashboardsDashboards turn a workflow's data into shareable visual summaries. API · formsA Sheetgo form collects data straight into a workflow. API · webhookA webhook is a private URL that lets external systems push data into a workflow over HTTP. API · api-destinationAPI as a destination turns a workflow's output into a live, read-only JSON feed at a private URL. API · integrationsIntegrations are pre-configured sources for third-party services — currently Shopify and QuickBooks. API · outlookIntegration endpoints used by the Sheetgo add-in for Microsoft Outlook to connect mailbox data to workflows. API · gmailSend email through a connected Google account. API · filesBrowse and search the storage providers connected to your account — Google Drive, OneDrive, SharePoint and Dropbox — to find the ids you need when… API · bigqueryBigQuery can be both a source and a destination for an automation: read a table or a query's results into a spreadsheet, or load a spreadsheet's rows… API · authOAuth2 authorization endpoints — exchange a Client ID / Client Secret for an access token, for developer and customer integrations. API · ModelsAll 147 request and response models, field by field. API · GuidesRate limits, API key scopes, smart tags, cell references. Python examplesComplete Python recipes: errors and retries, pagination, create and run an automation, column letters, MCP tool calls, an AI agent with the Sheetgo tools.
Read this first if you are an AI agent

For AI agents and LLMs

How these pages are organised. This site is plain, server-rendered HTML, one page per topic. Every section is in the page source, even the ones shown collapsed. A plain-text index of all pages is at llms.txt.

  • index.html (this page): choosing between the API and MCP, the API key, scopes, glossary.
  • mcp.html: everything about the MCP server, including every tool with its parameters, return values and errors.
  • api.html: REST API overview (auth, conventions, errors, quickstart) and the index of every endpoint.
  • api-<group>.html: every operation of one group in full (parameters, request body, responses, errors), plus every model those operations use, so one page is enough to build a request.
  • api-models.html: all models in one place. api-guides.html: rate limits, key scopes, smart tags, cell references.
  1. Choose the REST API or MCP with the table in API or MCP?. If the user did not say, ask what they want to do.
  2. For the REST API: open the endpoint index, then the group page of the endpoint you need. Read the operation and the models listed under "Models used on this page" before you write the request. These pages are a snapshot from 2026-11-04 17:53 UTC; the live, latest version is at https://dev-api.sheetgo.com/rest/llms.txt and https://dev-api.sheetgo.com/rest/reference.md (Markdown), and the spec at https://dev-api.sheetgo.com/rest/openapi.json. If they disagree, the live docs win.
  3. For MCP: the tool reference matches the server's tools/list at the time this page was generated. Once connected, the server's own tools/list is the source of truth.
  4. Do not guess. Do not invent endpoints, fields, enum values or tool parameters that are not shown in these docs. If something is missing, fetch the linked page or say that you do not know.
  5. Handle the API key as a secret. Ask the user for their sg_… key, send it only in the Authorization header, and never print it back or store it in files you share.
  6. Ask before destructive actions. Deleting files, clearing ranges, deleting tabs, archiving workflows and running automations change the user's data. Confirm first, and use dry_run where a tool offers it.

The /rest pages (Swagger, ReDoc) need JavaScript for people, but they also contain a server-rendered text summary, and every reference page has a Markdown copy: add .md to the URL, or send Accept: text/markdown.

Decide

REST API or MCP?

They do different jobs. The REST API controls Sheetgo (the automations that move data between your files on a schedule). The MCP server lets an assistant work inside your Google files right now, in a conversation. Many projects use both.

REST APIMCP server
What it acts onSheetgo objects: workflows, automations (with their sources, processors and destinations), triggers and schedules, forms and their responses and approvals, dashboards, sharing, run status. Also file discovery, BigQuery discovery and sending email.Your Google files, directly: read and write Google Sheets, browse and organise Google Drive, read and fill Google Docs, read PDFs (text, tables, OCR), export and merge PDFs, run read-only BigQuery queries.
Typical callerAny program that speaks HTTP: scripts, back ends, no-code tools, CI jobs, and agents with an HTTP tool.An AI assistant or agent that supports MCP: Claude Code, Claude Desktop (through a bridge), Cursor, VS Code, or your own MCP client.
ProtocolHTTPS + JSON, described by OpenAPI 3.Model Context Protocol (JSON-RPC 2.0) over Streamable HTTP: tools/list and tools/call.
TimingRunning an automation is asynchronous: POST …/run answers 200 when the run is queued; you then poll its run status.Synchronous: every tool call does its work and returns the result.
Storage coveredGoogle Drive and Sheets, Microsoft OneDrive and SharePoint (Excel), Dropbox, BigQuery, Outlook, Gmail, APIs and webhooks, Shopify and QuickBooks.Google only: Drive, Sheets, Docs, PDFs and images stored in Drive, BigQuery.
Works without Sheetgo running anythingNo: automations run on Sheetgo's servers, on demand or on a trigger, even when nobody is online.Yes: each call happens right away and nothing keeps running afterwards.
AuthenticationSheetgo API key (Authorization: Bearer sg_…) or OAuth2.Sheetgo API key only (Authorization: Bearer sg_… or x-api-key: sg_…).
LimitsPer account: 120 reads and 30 writes per minute (read the X-RateLimit-* headers). Rate Limiting guide.Per API key: 120 calls per minute. Heavy file tools (PDF, OCR, export, merge, versions) run one at a time per key.
DocumentationSwagger, ReDoc, Markdown pages, llms.txt, OpenAPI JSON.This page, plus tools/list from the server.

Rules of thumb

  • Use the REST API to create, change, run or schedule Sheetgo workflows and automations; to keep data moving between files on a schedule; to work with Microsoft, Dropbox, Shopify, QuickBooks or webhooks; or to build a back-end integration.
  • Use MCP when an assistant should read, analyse, write or tidy up Google files during a conversation: "summarise this sheet", "append these rows", "fill this Doc template and export it to PDF", "find the latest invoice PDF and extract its table".
  • Use both when an assistant prepares files with MCP and then sets up a recurring Sheetgo automation with the API. Example: MCP checks the {{tags}} in a Docs template with docs_list_tags, then the API creates an automation with a document destination that merges every row into that template.
Both products

Before you start

  1. A Sheetgo account. Sign in at https://app.sheetgo.com. API calls and MCP tools act as you: they can reach exactly what your account can reach.
  2. Connect your storage in Sheetgo. Sheetgo uses the connections you already authorised in the web app. MCP needs your Google account connected (it is when you sign in with Google). BigQuery needs BigQuery authorised in the web app. The API's Microsoft, Dropbox, Outlook and integration features need those accounts connected.
  3. Create an API key. In Sheetgo open Account → Services and account → Sheetgo API (https://app.sheetgo.com/account) and create a key. It starts with sg_ and is shown only once, so copy it straight away. You can create several keys (for example one per tool or agent) and revoke any of them.
  4. Optionally, limit the key. A new key has full access. Scopes can limit it to some actions (view, edit, delete) and some items. Give an AI agent the smallest scope that does the job, for example View everything. See MCP permissions and the API Key Scopes guide.

Keep the key secret. Anyone with the key can act as you. Store it in an environment variable or a secret manager, not in code, prompts or shared files. If a key leaks, revoke it in the Account page.

Terms

Glossary

WorkflowA Sheetgo project that groups automations and runs them together, on demand or on a trigger.
Automation (also connection)One data transfer inside a workflow: one or more sources, optional processors (filter, merge, AI, anonymize…) and one or more destinations.
TriggerWhat runs a workflow automatically: a schedule, or a change to a file.
ProviderWhere a file lives: drive (Google Drive and Google Sheets), onedrive, sharepoint, dropbox.
Smart tagA {{Header}} placeholder that Sheetgo fills with a value from your data (documents, emails, file names).
API keyA secret starting with sg_ that acts as your Sheetgo user. Used by both the REST API and MCP.
ScopeA limit on what an API key may do: actions (view, edit, delete) on resources (workflows, automations, forms, sheets, drive, bigquery…).
MCP toolOne operation the MCP server offers, such as sheets_get_values. Listed by tools/list, run by tools/call.